1. Field of the Invention
The present invention relates to a content retrieval apparatus and method for retrieving a content description in which described are contents of a multimedia content (hereinafter referred to as a content) such as a video, audio and text.
2. Description of the Related Art
MPEG-7 has been standardized that is a description interface for retrieving and filtering a multimedia content including a video, image and audio which cannot be retrieved conventionally. It is thereby made possible to create various content descriptions using description schemes based on the XML schema standardized by W3C. Thus, the use of MPEG-7 is ready to start in broadcast, cellular communications, internet, etc.
The processing of content retrieval using MPEG-7 will be described below. MPEG-7 describes viewpoints (keywords) used in retrieving a content for each content description (for example, Japanese Laid-Open Patent Publication 2001-160066). When a client enters a desired term (keyword) as a request query for a content description, the term used in the request query entered by the client is compared with the viewpoint described in each of content descriptions stored in advance. Then, the content description is retrieved in which described is the same viewpoint as the request query. The retrieval of the content description which the client requests is thus performed.
In MPEG-7, a content description can be assigned a plurality of viewpoints. It is thus possible to assign a plurality of viewpoints described in different languages to a content description, and therefore, when a client performs content retrieval, the client is capable of using terms expressed in different languages, as a term for use in a request query. For example, when entering a name of Yamada as a term for a request query, it is possible to enter “Yamada” in English or “◯◯” in kanji character in Japanese. In other words, MPEG-7 supports multilingual retrieval function. In addition, “◯◯” are assumed to be kanji characters in Japanese expressing “Yamada” in English.
Further, in MPEG-7, a content description can be assigned a plurality of different terms having the same meaning, as viewpoints. It is thus possible for a client to use different terms having the same meaning as a term for use in a request query in performing content retrieval. For example, when entering a name of YAMADA as a request query, it is possible to enter “Yamada” in English, “Yama” as a nickname, “◯◯” in kanji character in Japanese, or “ΔΔΔ” in hiragana character in Japanese. In addition, “ΔΔΔ” is assumed to be hiragana characters in Japanese expressing “Yamada” in English.
However, in MPEG-7, when the number of corresponding languages is increased, it is necessary to describe in a content a plurality of viewpoints described in different languages. In other words, in MPEG-7, as the number of corresponding languages is increased, the data amount of the content description to create is increased.
Similarly, in MPEG-7, when the number of corresponding terms (vocabularies) is increased, it is necessary to describe in a content a plurality of terms viewpoints having the same meaning. In other words, in MPEG-7, as the number of corresponding terms is increased, the data amount of the content description to create is increased.
As described above, in MPEG-7, the number of corresponding languages and the number of corresponding terms are increased, whereby the number of expressions for term is increased and the degree of redundancy of content is increased. As a result, the cost performance deteriorates due to increases in the number of expressions for term, and it is made difficult to manage contents.